Station platform monitoring system

Abstract

To provide a station platform monitoring system which can accurately detect a dangerous event on a station platform and evoke a train driver's attention.SOLUTION: A station platform monitoring system which detects and notifies dangerous events on a station platform based on images taken by cameras installed on a vehicle comprises plural cameras installed on side surfaces of the vehicle, dangerous event detection means provided so as to correspond to each of the plural cameras, for analyzing the taken images and detecting the dangerous events, determination means for logically calculating detection results by plural dangerous event detection means provided so as to correspond to the cameras adjacently installed and determining generation domains of the dangerous events, and detection results output means for notifying generations of the dangerous events to a perception based on the determination results by the determination means.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to a surveillance system for station platforms, and more particularly to a surveillance system that processes images taken by cameras installed on the sides of train cars to monitor safety on station platforms. [Background technology]

[0002] Conventionally, safety checks on station platforms when trains depart have generally been carried out by constructing a system that visually informs crew of the status of passengers on the station platform using several cameras installed above the platform and a display device that shows images captured by these cameras. Additionally, an invention has been proposed for a platform monitoring device that divides the station platform into a vehicle section and a platform section based on images from a camera installed above the platform to capture the edge of the platform and the track section, determines the status of trains entering / departing from the image of the vehicle section, and detects whether there are people in the area to be monitored from the image of the platform section, and issues an alarm signal if people are detected (see Patent Document 1).

[0003] However, in the above-mentioned system that notifies users by using camera images, there is a problem that the monitored target may be lost when there are overlapping people or when a user is hiding behind an object. In response to this, an invention has been proposed for an image processing device in which cameras are installed not on the station platform but on the side of the train, directly above the doors, so that the outside of the train is captured in the shooting range, and images from adjacent cameras are stitched together to reduce blind spots, and a transparency is set in each specified area of ​​the images so that people, etc. captured in the seams do not disappear, and the first and second images are synthesized into a continuous image (see Patent Document 2). [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 10-338135 [Patent Document 2] JP 2019-191853 A Summary of the Invention [Problem to be solved by the invention]

[0005] The platform monitoring device described in Patent Document 1 has the problem of losing the target of monitoring when people overlap or when users are hiding in the shadows, and detects whether there is a person in the area of ​​the platform that needs to be monitored by comparing an image taken and saved when platform monitoring starts with an image taken when a train approaches and departs. However, the shooting environment (especially brightness) of a station platform changes significantly depending on the weather and time of day. Therefore, the platform monitoring device of Patent Document 1, which detects by comparing with images taken and saved in the past, has the problem of being unable to perform accurate detection because it detects by comparing images taken under different conditions.

[0006] On the other hand, the image processing device described in Patent Document 2 can prevent people and the like captured at the seams of images from adjacent cameras from disappearing, but since the driver looks ahead when starting the train and takes his eyes off the display screen, there is a risk that a user who has come into contact with the vehicle or been caught in the door may be overlooked even if he is captured in the image. In addition, the image processing device in Patent Document 2 is premised on the final decision being made by a person, so there is a problem that misrecognition or oversight may occur.

[0007] The present invention has been made in response to the above-mentioned problems, and aims to provide a station platform monitoring system that can accurately detect dangerous events on station platforms and thereby alert train drivers. Another object of the present invention is to provide a station platform monitoring system that can avoid the occurrence of dangerous situations due to driver misidentification or oversight on the display screen. Yet another object of the present invention is to provide a monitoring system for a station platform that can detect dangerous situations on the platform and issue appropriate warnings even if the shooting conditions of the surveillance cameras on the station platform change significantly. [Means for solving the problem]

[0008] In order to achieve the above object, the present invention provides A station platform monitoring system that detects and notifies dangerous events on a station platform based on images captured by a camera mounted on a vehicle, Multiple cameras installed on the sides of the vehicle; a risky event detection means provided corresponding to each of the plurality of cameras and configured to analyze captured images to detect risky events; a determination means for determining an occurrence area of ​​a dangerous event by performing a logical operation on detection results by the plurality of dangerous event detection means provided corresponding to adjacently installed cameras; a detection result output means for notifying a sense of occurrence of a dangerous event based on a result of the determination by the determination means; A driving information acquisition means for acquiring information regarding driving of a vehicle to be monitored; an image quality evaluation means for evaluating the image quality of an image captured by the camera; Equipped with 、 the determination means narrows down the detection results of the corresponding dangerous event detection means in accordance with the evaluation result by the image quality evaluation means, The detection result output means is configured to change the intensity of the notification depending on the driving information acquired by the driving information acquisition means. This is what was done.

[0009] According to the station platform monitoring system having the above configuration, the occurrence of a dangerous event is determined by logically operating the detection results by the multiple dangerous event detection means, so that erroneous detection or non-detection of dangerous events on the station platform can be prevented and dangerous events can be accurately detected. In addition, the occurrence of a detected dangerous event is notified to the senses by the detection result output means, so that the occurrence of a dangerous situation due to misrecognition or oversight of the display device screen can be avoided. Furthermore, since the detection results of the corresponding dangerous event detection means are narrowed down according to the evaluation result by the image quality evaluation means, the occurrence of a dangerous event can be detected even if the image quality of the captured image of any one of the multiple cameras is degraded due to the influence of the shooting conditions such as weather.

[0010] In addition, a driving information acquisition means is provided for acquiring information regarding the driving of the monitored vehicle, and the detection result output means changes the intensity of the notification depending on the driving information acquired by the driving information acquisition means. Therefore, for example, the intensity of the notification for the occurrence of a dangerous event changes when the vehicle is moving and when the vehicle is stopped, so that the driver can be sure to recognize the occurrence of a dangerous event even if he or she is looking straight ahead when setting off.

[0011] Further, it is preferable to provide a detection condition changing means for changing a detection condition by the hazardous event detection means in accordance with the driving information acquired by the driving information acquisition means. According to this configuration, the conditions for detecting a dangerous event change depending on, for example, whether the vehicle is moving or stopped, so that the occurrence of a dangerous event can be accurately detected according to the situation.

[0012] Also, preferably, the multiple cameras installed on the sides of the vehicle include a camera installed on the side of the front end of the vehicle so as to photograph the center of the vehicle, and a camera installed on the side of the rear end of the vehicle so as to photograph the center of the vehicle. With this configuration, even if passengers getting on and off the platform are captured overlapping in the image from one camera, hidden passengers will be captured in the image from the other camera, preventing dangerous events from going undetected.

[0013] The detection area of ​​the monitored object may be divided into a plurality of areas set in stages according to the degree of danger, and the detection result output means may report the occurrence of a hazardous event with a reporting intensity according to the degree of danger of the area in which the occurrence of the hazardous event is detected by the determination means. According to this configuration, the detection result output means reports the occurrence of a hazardous event with a reporting intensity according to the degree of danger of the area in which the occurrence of the hazardous event is detected by the determination means, so that the driver can estimate the degree of danger from the reporting intensity.

[0014] Also, preferably, the vehicle includes a weather information acquisition means for acquiring information regarding the weather when the vehicle is being driven, The driving information acquired by the driving information acquisition means includes vehicle travel direction information, The image quality evaluation means evaluates the image quality of the image captured by the camera in accordance with the vehicle travel direction information acquired by the driving information acquisition means and the weather information acquired by the weather information acquisition means. According to this configuration, the image quality of the image captured by the camera is evaluated according to the vehicle's direction of travel and the weather, so that a deterioration in the image quality can be determined without performing tedious and time-consuming image processing, thereby enabling the occurrence of a dangerous event to be accurately detected. Effect of the Invention

[0015] According to the station platform monitoring system of the present invention, it is possible to accurately detect dangerous events on the station platform, thereby alerting the train driver. It is also possible to avoid dangerous situations caused by misidentification or oversight of the display screen. Furthermore, it is possible to detect dangerous events on the platform even if the shooting conditions of the surveillance cameras on the station platform change significantly. [Brief description of the drawings]

[0016] [Figure 1]1A and 1B show an example of how to install a camera for acquiring image data used in the station platform monitoring system according to the present invention, where (A) is a plan view of a vehicle and (B) is a side view of the vehicle. [Diagram 2] 1 is a block diagram showing an embodiment of a station platform monitoring system according to the present invention; [Diagram 3] 1A to 1D are diagrams showing examples of setting detection areas in a surveillance system according to an embodiment of the present invention. [Figure 4] 4A and 4B are block diagrams showing an embodiment of a dangerous event determination means constituting the monitoring system of the embodiment. [Diagram 5] 1 is a block diagram showing an embodiment of an integrated determination means constituting a station platform monitoring system according to the present invention; [Figure 6] 7 is a diagram for explaining a method for narrowing down detection results in the integrated determination means of FIG. 5. [Figure 7] 13 is a diagram for explaining another method for narrowing down the detection results in the integrated determination means. FIG. [Figure 8] FIG. 11 is a block diagram showing another embodiment of the station platform monitoring system according to the present invention. [Figure 9] 13A and 13B are diagrams showing an example of setting a detection area in a modified example of the monitoring system of the embodiment. [Figure 10] 13A and 13B are diagrams showing an example of how to install a camera and set a detection area in a second modified example of a surveillance system according to the embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0017] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, preferred embodiments of a station platform monitoring system according to the present invention will be described in detail with reference to the drawings. Fig. 1 shows an example of the mounting positions of cameras that capture image data used in the station platform monitoring system according to the present invention, where (A) is a plan view of the vehicle and (B) is a side view of the vehicle. Although not shown, the same applies to other vehicles that make up the same formation.

[0018] 1, in the station platform monitoring system of this embodiment, cameras 11A(R), 11A(L) are installed at the front ends of the left and right sides of each vehicle 10, with angles set so that they can capture images of the side of the vehicle and the platform from the front toward the center of the vehicle. Also, cameras 11B(R), 11B(L) are installed at the rear ends of the vehicles, with angles set so that they can capture images of the side of the vehicle and the platform from the rear toward the center of the vehicle. In Fig. 1(A), SA1 is the horizontal shooting range of camera 11A(L), SA2 is the horizontal shooting range of camera 11A(R), SA3 is the horizontal shooting range of camera 11B(L), and SA4 is the horizontal shooting range of camera 11B(R). Also, in Fig. 1(B), the arrows indicate the shooting directions, θa is the vertical shooting range of camera 11A(R), and θb is the vertical shooting range of camera 11B(R). The same is true for cameras 11A(L) and 11B(L) on the opposite side.

[0019] A block diagram of an embodiment of a station platform monitoring system according to the present invention is shown in Fig. 2. This monitoring system is provided for each train set. However, detection may be performed for each car that constitutes each train set. 2, the monitoring system of this embodiment includes a dangerous event detection means 12A that receives and digitally processes images captured by the camera 11A(R) or 11A(L) on the front side of the vehicle to detect the occurrence of dangerous objects or dangerous events, and a dangerous event detection means 12B that receives and digitally processes images captured by the cameras 11B(R) and 11B(L) on the rear side of the vehicle to detect the occurrence of dangerous objects or dangerous events. The cameras 11A(R) to 11B(L) are preferably video cameras capable of capturing moving images, but may also be still cameras capable of continuously capturing multiple still images in a short period of time.

[0020] In addition, the monitoring system of this embodiment is equipped with a driving information acquisition means 14 that acquires driving information such as driving speed information and traveling direction information from an information management device 13 that is mounted on the vehicle 10 and collects various information related to driving, and the above-mentioned dangerous event detection means 12A, 12B are configured to detect dangerous objects and dangerous events based on the video information from the camera and the driving information acquired by the driving information acquisition means 14. Furthermore, the monitoring system of this embodiment includes a detection condition changing means 15 that changes the danger detection conditions based on the driving information acquired by the driving information acquiring means 14 .

[0021] The detection condition change means 15, for example, determines the next station to stop at based on operation information, judges the platform configuration of the station, and changes the detection target area and detection method according to the platform configuration (for example, whether it is straight or curved).The detection condition change means 15 also changes the detection area for each vehicle based on train position information, distinguishing between vehicles approaching the platform and those not approaching it when entering the platform, or between vehicles that have left the platform and those that have not left the platform when departing. Furthermore, if there is a staircase at the platform, it is expected that many passengers will rush to board the train, so the area near the staircase is set as the detection target area even if it is far from the tracks. Note that by acquiring information on the degree of congestion on the station platform or time information as operation information, the detection area may be expanded during busy times and narrowed during less crowded times.

[0022] The monitoring system of this embodiment also includes an integrated judgment means 16 that performs an integrated judgment based on the detection results of the dangerous event detection means 12A and 12B, and a detection result output means 17 that notifies the occurrence of a dangerous event based on the detection result judged by the integrated judgment means 16, for example, by audio from a speaker installed in the driver's seat or by an image on a display device. The integrated judgment means 16 also performs a similar judgment on images from cameras installed in other vehicles. The dangerous event detection means 12A and 12B, the driving information acquisition means 14, the detection condition change means 15, and the integrated judgment means 16 can be realized by the cooperation of a microprocessor (MPU) and a program executed by the MPU.

[0023] The integrated judgment means 16 is configured, for example, by a logical operation means that performs a logical operation (logical OR or logical AND) between the detection results of the dangerous event detection means 12A and the detection results of the dangerous event detection means 12B, and can be configured so that it can be set from the outside whether the logical OR or logical AND operation is to be performed. For example, if you want to reduce the number of undetected images, you can select the logical sum operation, and if you want to reduce the number of false positives, you can select the logical product operation. The image from the camera is affected by the weather, and when the weather is good, a clear image is obtained, and when the weather is bad, an unclear image is obtained. Therefore, a sensor capable of detecting the weather, such as a light meter, rain gauge, or hygrometer, may be provided in the vehicle, and according to the signal from the sensor, when it is sunny, the logical sum operation may be automatically selected to reduce the number of undetected images, and when it is cloudy or rainy, the logical product operation may be selected to reduce the number of false positives.

[0024] Also, although not particularly limited, in the monitoring system of this embodiment, the detection result output means 17 is set so that the closer the area in which the hazardous event is detected to the vehicle, the higher the notification intensity (strength of the alert) to be output. Furthermore, the faster the vehicle speed is at the timing of detecting the hazard, the higher the notification intensity becomes, and when the hazard is detected while the vehicle is stopped, the notification intensity is set to be lower. Note that the signal (information) supplied from the detection condition changing means 15 to the hazardous event detecting means 12A, 12B may also be supplied to the detection result output means 17, so that the notification intensity is changed in response to the change in the detection conditions.

[0025] The change in the notification intensity may be stepwise or stepless depending on the distance from the train. In addition, in a platform where a fence or gate is provided, when a danger is detected outside the fence or gate (the side closer to the track), the notification intensity may be set higher than when a danger is detected inside. In the above embodiment, the determination is based on the image of the camera mounted on the train, but the train position may be determined by combining images from both the camera mounted on the train and the camera installed on the platform, information on the number of trains held by the information management device 13, and information that provides absolute position information such as signal message information transmitted from the ground, or the train position may be determined from the presence or absence of reflected information of a signal irradiated to the platform from a signal irradiating device attached to the end of the train. In the case of a system that uses images from both the camera of the train and the platform for determination, after departure, an alert may be issued if a dangerous situation is detected on the traveling side of the rear end of the train, and an alert may not be issued if a dangerous situation is detected in an area where the rear end of the train has passed.

[0026] Furthermore, the detection result output means 17 of this embodiment is equipped with a display device, and as shown in Fig. 3(A), the detection areas A1, A2, A3, and A4 projected onto the platform are displayed in different colors, such as red, yellow, orange, and green, according to the level of danger, making it easy to understand in which area a dangerous event is occurring. Here, the detection area A1 is a danger area, A2 is a caution area, A3 is a caution area, and A4 is a safe area. Instead of displaying the areas in stages as in Fig. 3(A), the level of danger may be displayed by gradually changing the color without any steps.

[0027] Furthermore, the video portion of the detected event is displayed surrounded by a detection frame, and the color of the frame is changed to make it easier to understand the level of danger. Figure 3 (B) shows an example of an area display for a platform with a fence F, and (C) shows an example of an area display for a platform while a train is running. In the case of a platform with a fence F, the outside of the fence F (the side facing the tracks) can be designated as a danger area A1, and the inside of the fence F as a safety area A4. When logical sum is selected as the logical operation by the integrated judgment means 16, the detection result output means 17 may display a detection frame surrounding the dangerous event detection area in the image from the camera that has not detected the danger by converting the coordinates of the detection frame displayed in the image from the camera that has detected the danger. Also, in the above embodiment, dangerous event detection means 12A, 12B are provided corresponding to the two cameras, respectively, and judgment is made based on the images from the two cameras, but the present invention can also be applied to a case where three or more cameras and corresponding dangerous event detection means are provided, and judgment is made based on the images from the three or more cameras.

[0028] Next, the details of the functions of each means constituting the monitoring system of this embodiment will be described with reference to Figures 4 to 6. Figure 4 shows a more detailed configuration example of each of the risky event detection means 12A and 12B. The dangerous event detection means 12A shown in Fig. 4(A) is configured with a detection area setting means 21 that sets a detection area for an image captured by a camera based on detection area information set for each platform of each station and stored in a storage device and vehicle operation information (traveling speed, vehicle position, etc.), a dangerous event determination means 22 that analyzes the captured image to detect dangerous events such as a fall or rushing to board a train, and a dangerous event region determination means 23 that narrows down the dangerous events within the detection area from among the detected dangerous events. The dangerous event detection means 12B has a similar configuration.

[0029] The detection area set by the detection area setting means 21 is a three-dimensional space defined by the distance from the side of the vehicle, the distance from the camera in a direction parallel to the vehicle, and the height from the platform, and is set in stages such as danger area, caution area, caution area, and safe area, in order from the side closest to the vehicle. In addition, targets of dangerous events include people, as well as animals, flying objects, baby strollers, wheelchairs, suitcases, white canes, and other objects that may come into contact with the vehicle or be caught in the vehicle door. In addition, various algorithms and DNNs (deep neural networks) using artificial intelligence have already been developed for detecting dangerous events by analyzing captured images, so such known technologies can be used.

[0030] The dangerous event detection means 12A shown in FIG. 4(B) is configured by further adding an image quality evaluation means 24 and an external information acquisition means 25 for determining whether the image quality of the images from each camera is sufficient for judging the detection of a dangerous event, to the dangerous event detection means 12A shown in FIG. 4(A), and by supplying the evaluation result by the image quality evaluation means 24 together with the video signal to the detection area setting means 21 and the dangerous event judgment means 22. The dangerous event determination means 22 determines the image quality by image processing, and if the image quality of either the front or rear camera is poor, detects a dangerous event mainly using the image from the camera with the better image quality.

[0031] Here, causes of degradation in image quality include weather-related factors such as rain or snow on the lens, and vehicle orientation-related factors such as backlighting. Degradation in image quality due to rain or snow can be grasped by performing frequency analysis on the image information from the camera to score the clarity. Degradation in image quality due to backlighting can be grasped by obtaining the distribution of pixel brightness values ​​of the image information from the camera and scoring the content of high brightness pixels for each camera. The image quality evaluation score calculated as described above may be configured to be output to the integrated judgment means 16 as described later and used for judgment.

[0032] In the dangerous event detection means 12A in Fig. 4(B), the dangerous event determination means 22 determines the image quality based on the video from the cameras, but a sensor such as a rain gauge may be provided in the vehicle to determine the cameras whose image quality is degraded by rain or snow and the extent of the degradation in image quality from the detection value of the sensor and information on the traveling direction of the vehicle acquired by the driving information acquisition means 14. Furthermore, the quality of image quality may be determined by determining whether any of the cameras is in a backlit state from time information, vehicle traveling direction and position information, and weather information acquired from outside the vehicle by the external information acquisition means 25. In addition, instead of outputting the image quality evaluation score (multi-value data) calculated by the image quality evaluation means 24 to the integrated judgment means 16, it may be supplied as a binary value or a continuous value (analog value) indicating whether or not the image quality evaluation score can be used to detect a dangerous event.

[0033] Next, a second embodiment of the station platform monitoring system of the present invention will be described. The second embodiment is configured so that the image quality evaluation score for each camera calculated by the image quality evaluation means 24 in FIG. 4B is output to the integrated judgment means 16 and used for judgment. As shown in Fig. 5, the integrated judgment means 16 of this embodiment includes detection result narrowing down means 61A, 61B for narrowing down detection results with high accuracy based on the detection results from the risky event region judgment means 23 of each of the risky event detection means 12A, 12B shown in Fig. 2 and the image quality evaluation score from the image quality evaluation means 24, and a logical operation means 62 for logically operating the detection results by the detection result narrowing down means 61A, 61B. The logical operation performed by the logical operation means 62 is a logical AND or a logical OR, similar to the above embodiment.

[0034] Here, a specific example of the narrowing down process by the detection result narrowing down means 61A, 61B will be described with reference to Fig. 6. Note that the image enclosed in the display frame is the surveillance target (person). When the image quality evaluation score provided by the image quality evaluation means 24 is a multi-value or analog value, the detection result narrowing means 61A, 61B first performs threshold processing on the provided image quality evaluation score, and as shown in FIG. 6(A), if conditions are good, it is possible to detect up to a maximum detection distance dmax corresponding to the distance from the camera to the farthest edge of the detectable range A, but the maximum detection distance dth at the time of image acquisition and the minimum detection pixel number Sth of the detection object M at the maximum detection distance dth are obtained. This is performed for both the front and rear camera images. The detection distance d is calculated based on a point determined by the detection frame, and as an example, it is possible to identify whether a person is in the danger area or the caution area by comparing the coordinate positions on the image of the center position of the feet and the danger area. By using the position of the feet, which is not affected by individual differences in height, the area determination performance can be improved. In addition, objects with fewer pixels than the minimum pixel number Sth in the image, in the above case Smin to Sth, are excluded from the detection determination target, and detection determination is performed for objects with more pixels than the minimum pixel number Sth. Furthermore, since dth and Sth can be converted into each other, the detection results can be narrowed down even if either one is used.

[0035] In addition, the detection result narrowing means 61A, 61B judges the reliability from the calculated maximum detection distances dth(front) and dth(rear). As a specific example, as shown in FIG. 6(B), when a point P of the maximum detection distance dth(front) and dth(rear) is plotted in an orthogonal coordinate system with the distance d(front) from the front camera on the horizontal axis and the distance d(rear) from the rear camera on the vertical axis, if the point P is above the straight line L passing through the maximum distance dmax of each axis, the detection judgment result is judged to be reliable and the detection judgment result is valid and supplied to the logical operation means 62. In addition, if the point P is below the straight line L passing through the maximum distance dmax of each axis, the detection judgment result may be judged to be unreliable and the driver may be notified of this by another means. Here, the maximum distance dmax is the distance from the camera to the opposite end of the vehicle (≒ vehicle length).

[0036] On the other hand, when the image quality evaluation score supplied from the image quality evaluation means 24 is binary, the detection result narrowing means 61A, 61B can include the entire detection range of one vehicle as the detectable target range in any of the cases (a), (b), and (c) shown in Figure 7(A). In (a) to (c) of FIG. 7(A), the symbol Sa denotes an image range in which a dangerous event can be detected by the front camera, and the symbol Sb denotes an image range in which a dangerous event can be detected by the rear camera.

[0037] In Figure 7(A), (a) represents a case where images from both the front and rear cameras can be used, (b) represents a case where only the image from the front camera can be used, and (c) represents a case where only the image from the rear camera can be used. Figure 7(B) shows the states corresponding to (a), (b), and (c) in (A) expressed in an orthogonal coordinate system of distances d (front) and d (rear) from the camera. Figure 7(B) shows that in all of (a), (b), and (c) in (A), the point corresponding to point P in Figure 6 is above line L passing through the maximum distance dmax of each axis, so that the entire detection range for one vehicle is the detectable target range.

[0038] The third embodiment is a surveillance system in which the maximum distances dmax (front) and dmax (rear) from each camera are the distances to the camera installed in the vehicle farthest from the camera, as shown in FIG. 8(A), and an example of the system block configuration in this case is shown in FIG. 8(B). As shown in Fig. 8(A), in this embodiment, information is supplied from a dangerous event detection means provided corresponding to the camera on the far side of an adjacent vehicle to the integrated judgment means 16 provided corresponding to each vehicle. With this configuration, the integrated judgment means 16 can detect dangerous events occurring near the joints between the vehicles. Also, as shown in Fig. 8(B), the detection result output means 17 includes a common display device DSP provided in the driver's seat or the like and multiple speakers SPK provided in each vehicle, and may report information obtained from the integrated judgment means 16 of each vehicle for each vehicle, or may report the information collectively for the entire train.

[0039] (Modification) Next, a first modified example of the above embodiment will be described. In the above embodiment, an example was described in which a detection area was set for each vehicle when the entire vehicle was facing the platform, and dangerous events were detected. However, as shown in Fig. 9(A) and (B), when a part of the vehicle is on the platform, a detection area DA is set corresponding to the part on the platform, and dangerous events are detected. Fig. 9(A) shows when the train is leaving the station, and (B) shows when the train is entering the station. Also, it may be configured so that the detection area setting can be switched ON / OFF for each vehicle.

[0040] FIG. 10 shows a second modification of the above embodiment. In the above embodiment, a pair of cameras is provided at the front and rear of each vehicle facing each other. In the second modification, as shown in FIG. 10(A), a camera 11 having a lens capable of capturing a wide range, such as a fisheye lens, is provided at the top of each door of the vehicle 10, and dangerous events are detected based on the images of each camera. In this case, the detection area may be set as concentric semicircular areas DA1 and DA2 centered on the camera, as shown in FIG. 10(B), or a technique is known for converting (expanding) an image of a camera having a fisheye lens into an original rectangular image, so that the image may be converted using such a technique to set a detection area (see FIG. 3) parallel to the platform edge, as in the above embodiment. In FIG. 10(B), the definitions of the distance d from the camera and the number of pixels S in this modification are also described.

[0041] Although the preferred embodiment of the present invention has been described above, the present invention is not limited to the above embodiment. For example, in the case of a vehicle without an information management device 13, the driving information acquisition means 14 may acquire driving information such as travel direction information from operation signals related to driving operation, or may acquire running speed information from signals from wheel rotation sensors instead of driving information. In addition, driving information may be acquired from a ground side device that manages the driving status of the train. In addition, the position where the camera is mounted is not limited to the rear end and the front end of all vehicles, and in a formation with a small number of coupled vehicles, it may be only the rear end of the rearmost vehicle and the front end of the first vehicle. In addition, in the above embodiment, the detection result output means 17 is equipped with a display device installed in the driver's seat to notify the driver, but it may also be linked to speakers and indicator lights installed in each car to directly notify passengers on the platform. [Explanation of symbols]

[0042] 10 Vehicles 11 Camera 12. Means of detecting dangerous events 13 Information management device 14 Means of obtaining driving information 15. Means for changing detection conditions 16 Integrated judgment means 17. Means of outputting detection results 21 Detection area setting means 22 Means for determining dangerous events 23. Means for determining risk area 24 Image quality assessment methods 25 External information acquisition means

Claims

1. A station platform monitoring system that detects and notifies dangerous events on a station platform based on images captured by a camera mounted on a vehicle, Multiple cameras installed on the sides of the vehicle; a risky event detection means provided corresponding to each of the plurality of cameras and configured to analyze captured images to detect a risky event; a determination means for determining an occurrence area of ​​a dangerous event by performing a logical operation on detection results by the plurality of dangerous event detection means provided corresponding to adjacently installed cameras; a detection result output means for notifying a sense of occurrence of a dangerous event based on a result of the determination by the determination means; A driving information acquisition means for acquiring information regarding driving of a vehicle to be monitored; an image quality evaluation means for evaluating the image quality of an image captured by the camera; Equipped with the determination means narrows down the detection results of the corresponding dangerous event detection means in accordance with the evaluation result by the image quality evaluation means, The station platform monitoring system is characterized in that the detection result output means is configured to change the intensity of the notification depending on the operation information acquired by the operation information acquisition means.

2. A weather information acquisition means for acquiring information regarding the weather when driving a vehicle, The driving information acquired by the driving information acquisition means includes vehicle travel direction information, The station platform monitoring system according to claim 1, characterized in that the image quality evaluation means evaluates the image quality of the image captured by the camera based on vehicle direction information acquired by the driving information acquisition means and weather information acquired by the weather information acquisition means.

3. The station platform monitoring system according to claim 1 or 2, further comprising a detection condition changing means for changing the detection conditions used by the hazardous event detection means in accordance with the operation information acquired by the operation information acquisition means.

4. A station platform monitoring system as described in any one of claims 1 to 3, characterized in that the multiple cameras installed on the sides of the vehicle include a camera installed on the side of the front end of the vehicle so as to photograph the center of the vehicle, and a camera installed on the side of the rear end of the vehicle so as to photograph the center of the vehicle.

5. The detection area to be monitored is divided into a plurality of areas set in stages according to the degree of danger, A station platform monitoring system as described in any one of claims 1 to 4, characterized in that the detection result output means notifies the occurrence of a dangerous event with an alarm intensity corresponding to the degree of danger in the area in which the occurrence of a dangerous event is detected by the determination means.

Images